How we test

A score means nothing without the method behind it. This page is that method. If an article ever departs from it, the article says so.

Games

1. Play on the platform that matters. We state which platform and which build we played, and for how long. A review written after four hours says four hours.
2. Score what matters for the genre. A puzzle game is not judged on combat. Each review lists the criteria it uses before it gives a number.
3. Separate taste from defect. "I did not enjoy it" and "it is broken" are different findings and are reported differently.
4. Name the player it suits. Every verdict ends with who should buy it and who should skip it.

Hardware

1. Measure, do not impress. Frame rates, 1 % lows, thermals, noise in dB, latency and battery life — measured, with the test bench stated.
2. Same conditions, every time. Comparable products are tested on the same bench, at the same settings, in the same room.
3. Report the cost of the number. Frames per watt and noise under load matter as much as peak performance.
4. State the sample. Whether a unit was bought, loaned or supplied is written in the article.

What we will not do

• Publish a benchmark, a price, a drop rate or a play time we did not verify.
• Let a commission, a review sample or an embargo change a verdict.
• Present a rumour as a fact, or bury the correction at the bottom of the page.
• Score a product we have not used.

Corrections

When we get something wrong we correct it in place, note what changed and when, and keep the original claim visible. Write to hello@gameselecto.com.

Dated data

Games change with patches and hardware changes with revisions and drivers. Every figure we publish is stamped with the version it was measured on. If an article carries no stamp, the figure is not version-dependent.